F-14 radar generates false targets in TWS/TWS cannot resolve multiple targets

[KenobiOrder]

I have been checking up periodically on the status of the F-14s radar ever since TWS auto released. Since then, the radar has not worked correctly, with significant issues with false contact generation. It has been gradually improving, but the radar is still in a state that dubiously useful in combat because it behaves very erratically.

 

I did some testing today and saw many strange things happen, but I had the presence of mind to record two of them as tracks.

 

Issue 1: The radar randomly generates a second contact not even in the vicinity of the first in TWS auto. Radar then gets confused and fails to operate satisfactorily. This the track titled "false target 14" The Su-27 is flanking to the left when suddenly a second contact appears significantly displaced in azimuth to the left and significantly further in range. The contact can be seen on the DDD, and it is after this that the radar goes bananas.

 

Issue 2: Radar cannot reliably detect semi close targets in TWS. This track is from the BVR mission on Persian gulf that comes with the game. We are not talking about targets that are right on top of each other wing tip to wing tip etc. This is aircraft at a typical formation distance. Strangely, when you enter RWS the radar has no issue detecting these targets. When you go back into TWS, it loses them. Very occasionally it will break out both targets very briefly, but I can see no pattern to when it does this.

 

 

false target 14.trk TWS cant resolve targets.trk

Edited August 5, 2021 by KenobiOrder

[nighthawk2174]

Yeah with TWS it also seems that when it does break them out something funky happens where the new contact is re-designated as the old one.  Resulting in the loss of the track of the first target.  Which is odd as it should just designate the new target as a new target, doesn't make much sense that it drops the old track.  It then seems to get confused, you loose the new track, and you get a bunch of false contacts, often with one shooting off into space with a ludicrous velocity.  Resulting in the loss of all tracks and missiles missing.

Edited August 6, 2021 by nighthawk2174

[near_blind]

2 hours ago, KenobiOrder said:

This is aircraft at a typical formation distance. Strangely, when you enter RWS the radar has no issue detecting these targets. When you go back into TWS, it loses them. Very occasionally it will break out both targets very briefly, but I can see no pattern to when it does this.

In broad terms, the radar doesn't see four aircraft in a formation, it sees four blobs close together. How clear it can refine down those blobs, and how much it can break them out depends on a bunch of stuff including the beam width, and the ability of the radar to process the gains. The rudimentary computers driving the AWG-9 limit it in this respect, and it struggles to break things out at medium to long range. 

In RWS the radar is showing what are more or less raw radar returns. As it sweeps over our collection of blobs, it generates a hit on your TID and DDD every time it gets a return it thinks is valid. It doesn't really care about anything else. In TWS, the radar is doing a lot of processing, trying to take the collection of blobs and fit them into a coherent track object with a speed and a position so that it can find it again the next time it sweeps through that area. If it didn't, then the radar would constantly look like what you're describing in your first track, and you couldn't guide missiles. As part of the processing its trying to find the centroid of the radar contact, and as part of that, it needs to decide if that's really one big blob, or four small blobs. If it can't break them out well enough, it sees them as a single contact. There isn't a real pattern to the the radar breaking out targets other than the closer you get, the better the radar can break individual targets out of a formation. 

[KenobiOrder]

8 minutes ago, near_blind said:

In broad terms, the radar doesn't see four aircraft in a formation, it sees four blobs close together. How clear it can refine down those blobs, and how much it can break them out depends on a bunch of stuff including the beam width, and the ability of the radar to process the gains. The rudimentary computers driving the AWG-9 limit it in this respect, and it struggles to break things out at medium to long range. 

In RWS the radar is showing what are more or less raw radar returns. As it sweeps over our collection of blobs, it generates a hit on your TID and DDD every time it gets a return it thinks is valid. It doesn't really care about anything else. In TWS, the radar is doing a lot of processing, trying to take the collection of blobs and fit them into a coherent track object with a speed and a position so that it can find it again the next time it sweeps through that area. If it didn't, then the radar would constantly look like what you're describing in your first track, and you couldn't guide missiles. As part of the processing its trying to find the centroid of the radar contact, and as part of that, it needs to decide if that's really one big blob, or four small blobs. If it can't break them out well enough, it sees them as a single contact. There isn't a real pattern to the the radar breaking out targets other than the closer you get, the better the radar can break individual targets out of a formation. 

The range is not a factor. Even a close range it cannot break the contacts out. The range would hypothetically matter, but it doesnt make a different in DCS F-14 in TWS.

 

This should not be a problem for TWS. Clearly the radar is capable of seeing both contacts in terms of its resolution cell, so the issue is with the TWS logic. TWS seems to be filtering out the second contact as a false alarm, which it should not be doing. Since it sees both contacts on the same scan, these should count a separate observations. Tracking gates should then be made for both, and on the next scan it should determine which is which based on the statistical distance of each contact from the center of the gate. Its possible it could accidentally get the contacts swapped, but it would still reduce them to two independent contacts.

 

What it appears to be doing is seeing both contacts on the same scan, and treating one of them as a false alarm.

 

 

Additionally we can clearly see that there is an issue with how the radar discards false contacts in other situations. This combined with the issues with closely spaced targets is probably causing all the issues were seeing with false targets and the TWS radar becoming clutters with rubbish. 

 

In the first track you can see on the DDD that the flanker is approaching the beam. When it seems to hit the notch, a second very high velocity target is generated. The radar decided to treat BOTH of these are contacts and this is why we get as many as 3 contacts on the TID at one point.

 

What should be happening (either or):

 

-The high speed target would likely be filtered out entirely or treated as a totally new contact since its statistical distance should be outside the predicted gate for the existing track file.

 

-Since the target entered the notch, even if the target was within the valid statistic distance of the gate, the contact should have been dropped. The radar should have stored the new contact in memory and then rejected it as a valid contact when it didnt appear on the next scan. This is because TWS systems dont just display everything they see immediately. If a new contact is found outside the tracking gate, the radar should wait at least one scan before displaying this contact as a track.

Edited August 6, 2021 by KenobiOrder

[nighthawk2174]

13 minutes ago, near_blind said:

In broad terms, the radar doesn't see four aircraft in a formation, it sees four blobs close together. How clear it can refine down those blobs, and how much it can break them out depends on a bunch of stuff including the beam width, and the ability of the radar to process the gains. The rudimentary computers driving the AWG-9 limit it in this respect, and it struggles to break things out at medium to long range. 

Yes I understand if there are four targets all within a certain range of each other the radar, without special modes, will be unable to resolve them as individual targets.  However what i've noticed is that it does break them out (although RWS is for some odd reason much better which is part of the OP's post I'm just adding an additional issue with this) it drops the original track and starts tracking the new target.  Even though it should just go ohh there's a new target here and create a new track.  Then it for no good reason looses this new target (X through the contact) and you get a new new false target moving at Mach 10 in a complete different direction (Which TWS auto will track on).  Then all the targets just vanish and you don't see anything for a while.

13 minutes ago, near_blind said:

In RWS the radar is showing what are more or less raw radar returns. As it sweeps over our collection of blobs, it generates a hit on your TID and DDD every time it gets a return it thinks is valid. It doesn't really care about anything else. In TWS, the radar is doing a lot of processing, trying to take the collection of blobs and fit them into a coherent track object with a speed and a position so that it can find it again the next time it sweeps through that area. If it didn't, then the radar would constantly look like what you're describing in your first track, and you couldn't guide missiles. As part of the processing its trying to find the centroid of the radar contact, and as part of that, it needs to decide if that's really one big blob, or four small blobs. If it can't break them out well enough, it sees them as a single contact. There isn't a real pattern to the the radar breaking out targets other than the closer you get, the better the radar can break individual targets out of a formation. 

For TWS there will be an initial detection phase where it acts the same way as RWS.  You get the additional info from passes after initial detection.  This is why TWS needs to try and get a sweep on the target every few seconds to work properly.  This initial detection phase (from which tracks are built) should not be any worse at splitting targets then RWS, from my understanding its not much different if at all.  

[near_blind]

55 minutes ago, KenobiOrder said:

The range is not a factor. Even a close range it cannot break the contacts out.

Watching the track, the starting range is in excess of what I would expect the DCS AWG-9 to be able to break out two contacts with 4000 feet of separation. As their range closes to you, their separation decreases and they maintain similar altitudes airspeed. This looks consistent with how HB seem to have modelled their radar, at least to me. 

 

 

40 minutes ago, nighthawk2174 said:

it drops the original track and starts tracking the new target.  Even though it should just go ohh there's a new target here and create a new track.  Then it for no good reason looses this new target (X through the contact) and you get a new new false target moving at Mach 10 in a complete different direction (Which TWS auto will track on).  Then all the targets just vanish and you don't see anything for a while.

I'm not arguing the track shedding behavior is correct.

 

 

48 minutes ago, nighthawk2174 said:

For TWS there will be an initial detection phase where it acts the same way as RWS.  You get the additional info from passes after initial detection.  This is why TWS needs to try and get a sweep on the target every few seconds to work properly.  This initial detection phase (from which tracks are built) should not be any worse at splitting targets then RWS, from my understanding its not much different if at all.  

I'm not so sure, but I have nothing to prove or disprove one way or another. 

[Callsign JoNay]

5 hours ago, KenobiOrder said:

The range is not a factor. Even a close range it cannot break the contacts out.

 

 

It can break them out if you get close enough. See below a video where I'm tracking two Su-27 that are in a combat spread with 1nm separation. TWS breaks them into individual contacts at 25nm. This new RADAR behavior was introduced in the same patch where the F-18 got it's RAID improvements, so I assumed it wasn't a bug but a feature.

 

 

 

 

[AH_Solid_Snake]

Im not sure issue #2 is a problem and not expected behavior. The TWS cells for the AWG-9 are fairly broad (5nm 4deg?) so at extended range the azimuth delta for targets in formation will go below that threshold.

 

There is not actually a "initial RWS phase" for TWS, in that it wont ever show you the original radar return and it will wait a minimum of two (or three?) sweeps to present anything on the TID, once you're in TWS the stuff on the TID are track files not returns, in this way TWS even on its first pass is more limited than RWS given that so long as you can visually pick out a distinct target on the display, and the displays are actually pretty precise instruments it doesn't seem crazy to me that RWS is better in that regard.

 

 

[nighthawk2174]

It has to wait as it needs to pass over the target multiple times in order to build the track file, I still don't see how this would result in RWS having a better ability to break out targets then TWS.  TWS operates by combining these multiple observations into a track file over time.  I have not found anything that states that the actual waveform or processing of each individual return is any different from RWS just that multiple observations over time are combined into one track file.  This is why LTWS on the hornet can be a thing, it doesn't update as quick but it can give you bandit heading information without needing to pass over the target every few seconds.

[KenobiOrder]

4 hours ago, AH_Solid_Snake said:

TWS cells for the AWG-9 are fairly broad (5nm 4deg?)

Where are you getting that figure? My expectation would be that the res cell of TWS would be the same as RWS, or sufficiently similar. 5nm sounds fantastically large, and I dont think that is the case in game either because I have certainly detected targets more closely spaced than 5nm.

[KenobiOrder]

I have continued to test and found some more strange behavior.

 

Something of note is that the velocity vector indicators do not seem to be working. According to the manual they should be showing the velocity vector of the target, but instead the move to the right if the target is closing faster and to the left if slower, not actually showing the direct of the target.

 

Second, it can be seen on the DDD that that radar in TWS can generally see two targets, but does not output them on the TID.

 

 

wacky behavior.trk wacky behaviorno notch.trk failure to guide.trk

[near_blind]

56 minutes ago, KenobiOrder said:

Something of note is that the velocity vector indicators do not seem to be working. According to the manual they should be showing the velocity vector of the target, but instead the move to the right if the target is closing faster and to the left if slower, not actually showing the direct of the target.

From the Manual:

 

Quote

 

Velocity vector emanating from center dot of tracks when velocity vector display is selected.

 

Vector direction represents track heading and length represents track speed so that the max indicated speed (1 800 knots) is 1.5 inches on the TID.

 

In TID ground stabilized mode the vector direction represents track true heading and the vector length represents track ground speed.

 

In TID aircraft stabilized and attack modes the vector direction represents track relative heading (to own aircraft) and the vector length represents track speed relative to own aircraft.

 

In A/C Stab and ATTK modes, the TID isn't showing their actual velocity vector, but their heading and speed relative to you. As for why the velocity vector is going all crazy, it's because TWS is on the threshold of being able to pick out the individual contacts, but isn't quite there yet. It's trying to average out the velocity of the two flankers, and that's leading to some funky vectors. 

As for the missiles. Neither missile in the first track look like it received an active signal from the radar. There was discussion of an issue where extrapolated tracks that have wandered too far from the game object that they're meant to represent would never receive an active signal. There was supposed to be a fix for it in the last open beta, but I've seen more missiles fail to go active after the patch then before. I think this is a bug. Maybe @IronMike knows what that fix actually entailed. 

Your second track, it looks like the missile went active, but it bit off on chaff. 

The third track, it looks like the first missile never went active on the extrapolated track again. I have no explanation for why that second missile never received an active signal, that definitely looks like a bug. Third missile made stupid missile decisions and bled all its speed. 

[KenobiOrder]

6 minutes ago, near_blind said:

because TWS is on the threshold of being able to pick out the individual contacts, but isn't quite there yet. It's trying to average out the velocity of the two flankers, and that's leading to some funky vectors. 

This should not happen in a tws system. If the radar picks up two contacts, it would not display any change till it determines based on the statistical difference from last track which is which. There also would be no averaging of the two different aircrafts velocity. 

 

The ddd is showing two contacts, so the radar should be able to create separate track files.

 

What it might be doing, which I think is an error, is disregarding one of the contacts as a false alarm. Then it occasionally gets one of the contacts confused with the other based on their different maneuvers.

 

It might get the contacts accidently swapped, but it should still be tracking two contacts. And it's odd that it's swapping them when all they are doing is flying a straight line.

 

[near_blind]

21 minutes ago, KenobiOrder said:

This should not happen in a tws system. If the radar picks up two contacts, it would not display any change till it determines based on the statistical difference from last track which is which. There also would be no averaging of the two different aircrafts velocity. 

 

The ddd is showing two contacts, so the radar should be able to create separate track files.

 

The DDD, like RWS, is going to show raw doppler rate returns: because it sees two velocities doesn't mean the two contacts are going to appear on the TID. However if you don't believe that there is additional processing going on as part of TWS, nothing I say is going to convince you this isn't intended behavior. If nothing else I'd submit a separate thread for those missiles not receiving their active signal in addition to this thread.

[captain_dalan]

1. When the target starts notching you, did you change the radar aspect from nose to flank?
2. TWS is not the expected radar mode for engaging small nimble targets. You are expected to engage them in PD or P-STT. 

[KenobiOrder]

26 minutes ago, near_blind said:

 

The DDD, like RWS, is going to show raw doppler rate returns: because it sees two velocities doesn't mean the two contacts are going to appear on the TID. However if you don't believe that there is additional processing going on as part of TWS, nothing I say is going to convince you this isn't intended behavior. If nothing else I'd submit a separate thread for those missiles not receiving their active signal in addition to this thread.

Of course there is additional processing. But there isn't any good reason that tws should not generate separate track files if the radar is discriminating two targets. It's not a problem if the correlation areas of each contact overlap. When the radar sees two hits on the same bar scan it should generate a tentative track in memory for both. If on the next scan (or however many the radar needs) it detects then again, it should create two separate track files.

[nighthawk2174]

4 hours ago, captain_dalan said:

1. When the target starts notching you, did you change the radar aspect from nose to flank?

What does this switch do and is it modeled?

 

[near_blind]

14 minutes ago, nighthawk2174 said:

What does this switch do and is it modeled?

Quote

The ASPECT switch (21) controls two different things depending on radar mode. In the pulse doppler search modes it controls the rate processing windows of the radar, NOSE sets 600 knots opening to 1,800 knots closing, BEAM sets 1,200 knots closing to 1,200 knots opening and TAIL sets 1,800 knots opening to 600 knots closing. In the short pulse STT modes the switch sets the system tracking mode to the corresponding echo edge or centroid to counteract countermeasures like chaff and specific jammer modes.

The PD functionality is modelled, the PSTT is functionality is not. 

[Kula66]

8 hours ago, near_blind said:

There was supposed to be a fix for it in the last open beta, but I've seen more missiles fail to go active after the patch then before. I think this is a bug. Maybe @IronMike knows what that fix actually entailed. 

Likewise, I've not seen any appreciable difference in 54s picking up dropped tracks since the last patch; it's happening depressingly often even with minimal change in target aspect.

[sLYFa]

12 hours ago, captain_dalan said:

1. When the target starts notching you, did you change the radar aspect from nose to flank?

Just to be clear, that switch only controls min/max closing velocity (i.e. doppler frequency shift) the radar operates at. It does not help remedy the "natural" shortcomings of a doppler radar, i.e. targets with close to zero relative velocity and/or targets beaming you can not be seen. 

[captain_dalan]

3 hours ago, sLYFa said:

Just to be clear, that switch only controls min/max closing velocity (i.e. doppler frequency shift) the radar operates at. It does not help remedy the "natural" shortcomings of a doppler radar, i.e. targets with close to zero relative velocity and/or targets beaming you can not be seen. 

Indeed, but it should make the tracks dropping as a result of too much closure (either way) less frequent. I don't have enough confirmed cases, but on one training run i did notice the radar would fail to build tracks until i switched it from nose to flanking. 

[sLYFa]

The nose setting should be best in any scenario. If something is moving away from you at more than 600kts its not a factor anyway. The 1200 max closure the beam setting gives you is likely to become a problem in head on engagement on the other hand. TLDR, leave it in nose. 

[captain_dalan]

10 hours ago, sLYFa said:

The nose setting should be best in any scenario. If something is moving away from you at more than 600kts its not a factor anyway. The 1200 max closure the beam setting gives you is likely to become a problem in head on engagement on the other hand. TLDR, leave it in nose. 

Will consider it and continue with experimentation when able. Not soon though, as my schedule is pretty cramped right now  

[NaCH]

** Not sure if related but the outcome is very similar **

 

Trying to understand if this issues is related or not, but could be possible that now in TWS the TID show again missiles ?

Since the last patch on update in AUG I see multiple TID detections but these are not represented in the DDD.

 

 

So far this is happening in Multiplayer. I was thinking that could be related datalink information from multiple sources or something else. 

 

Without a human rio the best way to discriminate valid contacts is asking to do a brief PD - STT  to Jesters. 

 

Let me know if I should create another tread and track file for this or keep it here. 

 

[TordinVarglund]

Bumping this because I'm having massive issues with lost tracks recently that I did not have ~2-3 months ago.

I don't know what exactly is going on, but it seems to have something to do with targets flying in relatively close formation (like .5nm or less).

 

All of this is with Jester in TWS-A, and applies both in single and multiplayer.

 

Best way to test this is to jump into the PG BVR and Syria BVR Instant Action missions, and observe the difference in track behavior. The Syria one seems to spawn the two MiG-29s with wider separation, and the radar is better at separating them and maintaining independent track files, and missiles correctly guide and go active. In the PG one, it might pick up both targets initially, but it will very quickly reporting lost tracks (Track Extrapolated symbology).